7652
Biochemistry 1997, 36, 7652-7663
Role of Tryptophan-63 of the Kringle 2 Domain of Tissue-Type Plasminogen Activator in Its Thermal Stability, Folding, and Ligand Binding Properties† Yuan Chang, Jaroslav Zajicek, and Francis J. Castellino* Department of Chemistry and Biochemistry, UniVersity of Notre Dame, Notre Dame, Indiana 46556 ReceiVed January 28, 1997; ReVised Manuscript ReceiVed April 14, 1997X
ABSTRACT:
Conservative (F and Y) and radical (H and S) mutations have been engineered at a rigidly conserved aromatic residue, W63, of the isolated recombinant kringle 2 domain of tissue-type plasminogen activator (r-K2tPA), an amino acid residue predicted from the X-ray crystal structure to be important in the ligand binding properties of this isolated protein domain. The variants were expressed in Pichia pastoris cells. The binding constants of -aminocaproic acid (EACA), 7-aminoheptanoic acid (7-AHpA), and trans-(aminomethyl)cyclohexanecarboxylic acid (AMCHA) to each of these mutant polypeptides were determined by titrations of the alterations in intrinsic fluorescence of the variant kringles with the ligands. As compared to wild-type r-K2tPA, increases in the Kd (dissociation) values of approximately 15-fold and 20-200-fold were found for the W63F and W63Y mutants, respectively, toward these three ligands. Neither the W63H nor the W63S variant interacted with these same ligands. Differential scanning calorimetric analyses were also performed on each of the peptides to determine whether the alterations affected the conformational stability of wtr-K2tPA. The data demonstrated that all of these mutants were thermally destabilized, possessing temperatures of maximum heat capacity (Tm) values that were 12-20 °C lower than that of wtr-K2tPA. Addition of EACA resulted in increases (≈12 °C) in the Tm values of r-[W63F]K2tPA and r-[W63Y]K2tPA, a result showing that EACA stabilized the native conformations adopted by these kringle domains. As expected from its greatly diminished binding to r-[W63H]K2tPA and r-[W63S]K2tPA, high concentrations of EACA had little effect on the Tm of thermal denaturation of these latter mutants. 1H-NMR analysis of the two aromatic mutant kringles was employed to assess their overall comparative folding properties. The high upfield chemical shifts (-0.98 ppm) of the CH3δ′ protons of L47, a major signal of proper kringle folding, were slightly lowered to -0.83 to -0.86 ppm in the cases of all of the mutants. This is due to alterations in the W25-L47 side-chain spatial orientations, possibly the result of slight conformational alterations that affect the distance relationships of these two amino acid side chains. Assignments of nearly all of the protons of the aromatic residues in the W63F and W63Y mutants were accomplished, and few additional differences from their wild-type counterpart were noted. Reactivities of the mutants against four different monoclonal antibodies directed to wtr-K2tPA revealed the possibility that some small local conformational alterations might have resulted from the residues that have replaced the W63. We conclude that W63 possesses an important direct role in the ligand binding properties of r-K2tPA. This residue also contributes significantly to the stability of the native conformation of this kringle domain and perhaps to maintenance of local conformations.
Kringles are structural motifs found in a number of proteins related to those involved in blood coagulation and clot dissolution. These modules contain approximately 80 amino acids and are covalently stabilized by three rigidly conserved disulfide bonds in a 1-6, 2-4, 3-5 pattern. These modules occur singly, as in uPA1 (Gunzler et al., 1982), and with great multiplicity, as in apo-Lp(a), where 11-41 copies have been found (McLean et al., 1987; Koschinsky et al., 1991; Kraft et al., 1992; Marcovina et al., 1996). Of the proteins involved in fibrinolysis, Pg contains five kringles (Sottrup-Jensen et al., 1978), one exists in uPA (Gunzler et al., 1982), and tPA possesses two such structural units (Pennica et al., 1983). The kringle modules in serine proteases of these types do not constitute part of their catalytic domains but principally † Supported by Grant HL-13423 from the National Institutes of Health and the Kleiderer/Pezold family endowed professorship (to F.J.C.). * To whom to address correspondence. Telephone: (219) 631-6456. Telefax: (219) 631-8017. X Abstract published in AdVance ACS Abstracts, June 15, 1997.
S0006-2960(97)00197-9 CCC: $14.00
function as recognition units for other proteins in solution and for components of cell surfaces. As examples, the kringle modules of Pg and Pm interact with platelets (Miles et al., 1988), peripheral blood cells (Miles et al., 1988), endothelial cells (Wu et al., 1992), and bacterial cells 1 Abbreviations: tPA, tissue-type plasminogen activator; K2 tPA, the kringle 2 region (residues C180-C261) of tissue-type plasminogen activator; Pg, human plasminogen; Pm, human plasmin; K1Pg, the kringle 1 region (residues C84-C162) of human plasminogen; K2Pg, the kringle 2 region (residues C166-C243) of human plasminogen; K3Pg, the kringle 3 region (residues C256-C333) of human plasminogen; K4Pg, the kringle 4 region (residues C358-C435) of human plasminogen; K5Pg, the kringle 5 region (residues C462-C541) of human plasminogen; uPA, urokinase-type plasminogen activator; KuPA, the kringle region (residues C50-C131) of human urokinase; BSA, bovine serum albumin; Lp(a), lipoprotein(a); MAb, monoclonal antibody; EACA, 6-aminohexanoic acid; 7-AHpA, 7-aminoheptanoic acid; t-AMCHA, trans-(aminomethyl)cyclohexanecarboxylic acid; DSC, differential scanning calorimetry; ELISA, enzyme-linked immunoadsorbent assay; C50-MAb, total concentration of monoclonal antibody necessary for binding to 50% of the molecules containing its epitope; TOF-MALDI-DE-MS, time-offlight matrix-assisted laser desorption ionization with delayed-extraction mass spectrometry; Tm, temperature of maximum change in heat capacity; TOCSY, total correlation spectroscopy; NOESY, nuclear overhauser enhancement spectroscopy.
© 1997 American Chemical Society
Tryptophan-63 of tPA Kringle 2 (Ullberg et al., 1990). The kringles of HPm bind to its plasma inhibitor, R2-antiplasmin (Sugiyama et al., 1988; Hortin et al., 1989), to its substrates, fibrinogen (Lucas et al., 1983) and fibrin (Thorsen et al., 1981; Lucas et al., 1983), to the plasma protein effectors, histidine-rich glycoprotein (Lijnen et al., 1980) and tetranectin (Clemmensen et al., 1986), and to cell-bound actin (Lind & Smith, 1991) and enolase (Miles et al., 1991). Further, the KuPA and K2tPA modules have been shown to mediate binding of tPA and uPA to their fast-acting plasma inhibitor, plasminogen activator inhibitor 1 (Wilhelm et al., 1990; Kaneko et al., 1991; Mimuro et al., 1992). In the case of uPA, a region of the molecule that includes its KuPA domain interacts with components of the cell matrix (Stephens et al., 1992a), and for tPA, its K2 region also interacts with fibrin (Ichinose et al., 1986; van Zonneveld et al., 1986; Wilhelm et al., 1990). Additionally, this module also participates in the binding of tPA to its endothelial cell receptor (Hajjar et al., 1994) and to other cell types (Bizik et al., 1993). In a final example, the Lp(a) kringles mediate binding of this protein to lowdensity lipoprotein (Frank et al., 1996) and to fibrin (Scanu & Edelstein, 1994). Some of the above interactions are inhibited by ω-amino acids, such as EACA and lysine, thus providing a molecular basis for the hypothesis that proteins containing carboxyterminal lysine residues are able to interact with proteins containing ω-amino acid binding kringles. The importance of this mode of protein-protein interaction is emphasized by the observation that digestion of virgin fibrin clots by plasmin, thereby exposing carboxyl-terminal lysine residues, enhances binding of Pg and tPA to these clots and, in this manner, stimulates clot lysis (Christensen, 1985; Tran-Thang et al., 1986). Such a mechanism also provides a means for plasma carboxypeptidase B-type enzymes to attenuate clot dissolution potential (Bajzar et al., 1995). Other proteins that contain carboxyl-terminal lysine residues (Hortin et al., 1988; Miles et al., 1991), or even internal lysine -amino groups with carboxylate groups of D and E side chains, appropriately spatially positioned to mimic free EACA, also interact with plasmin (Sugiyama et al., 1988). The kringle domains that are known to bind these types of ω-amino acid ligands, and thus probably involved in these interactions, are K1Pg (Lerch et al., 1980; Menhart et al., 1991), K2Pg (Marti et al., 1994), K4Pg, (Lerch et al., 1980; Rejante et al., 1991; McCance et al., 1994), K5Pg (Novokhatny et al., 1989; McCance et al., 1994), K2tPA (Cleary et al., 1989; De Serrano & Castellino, 1992a,b; De Serrano et al., 1992b), and several of the Lp(a) kringles (LoGrasso et al., 1994; Scanu et al., 1994; Klezovitch & Scanu, 1996; Mikol et al., 1996). These types of observations have stimulated detailed study of the nature of the binding of ω-amino acids to kringle domains. Such work has been facilitated by the successful large-scale expression of these modules in bacterial cells (Menhart et al., 1991; De Serrano & Castellino, 1992b; De Serrano et al., 1992a; Menhart et al., 1993; Marti et al., 1994; McCance et al., 1994; Chenivesse et al., 1996; Klezovitch & Scanu, 1996) and determination of the three-dimensional structures of several of these ligand-bound isolated domain units in the crystal state (de Vos et al., 1992; Mulichak et al., 1991; Arni et al., 1994; Mathews et al., 1996; Mikol et al., 1996) and in solution (Atkinson & Williams, 1990; Byeon et al., 1991; Li et al., 1994). Information forwarded from these structural studies was of predictive value in identifica-
Biochemistry, Vol. 36, No. 25, 1997 7653 tion of the amino acid side chains that interacted with the ligand, and site-directed mutagenesis investigations provided a great deal of direct information on this topic (De Serrano & Castellino, 1990, 1992a,b, 1993, 1994a,b; De Serrano et al., 1992a,b; Hoover et al., 1993; McCance et al., 1994). The amino acid positioned at W63 in K2tPA is invariant in all kringles. As revealed through X-ray crystal structures, this amino acid side chain is situated in such a manner as to interact with the hydrophobic regions of ω-amino acids, and NMR analysis shows that this side chain is perturbed by addition of these types of ligands. While these methods suggest that W63 functions in ligand binding, its relative importance in this regard has not been assessed by these previous studies. Specifically, only predictions of side-chain involvement in binding can be made from X-ray analysis, and whether perturbations of NMR resonances of the polypeptide by ligand are direct or indirect are not readily ascertained. However, these techniques, in combination with site-directed mutagenesis studies, are invaluable as a cluster of approaches to rigorously resolve these questions. The amino acid residue, W63, is strategically ideal for mutagenesis studies because it exists in a spatial location in the crystal structure that predicts its involvement in ligand binding, its 1H-NMR resonances are perturbed by ligand binding, and NOE connectivities with the ligands are observed. Additionally, the rigid conservation of this amino acid residue in both ligand binding and non ligand binding kringles also suggests that W63 might be a determinant of proper kringle folding and/or stability of the native conformation of these modules. In order to evaluate this broad range of possible functions of W63 in kringles, we have undertaken a site-directed mutagenesis-based study of the effects of replacements of W63 in K2tPA on the properties of this domain. The results of this study are reported herein. MATERIALS AND METHODS Materials. Bovine fXa was donated by Enzyme Research Laboratories, Inc. (South Bend, IN). Restriction endonucleases were obtained from Fisher Scientific Co. (Springfield, NJ). Recombinant Pfu DNA polymerase was purchased from Invitrogen (San Diego, CA). T4 DNA ligase was a product of New England Biolabs (Beverly, MA). Hybridoma cell lines expressing MAbs 364, 623, 663, and 700, all directed to wtr-K2tPA, were obtained from Genentech, Inc. (South San Francisco, CA). The Pichia pastoris expression kit and the yeast transfer vector, pPIC9K, were purchased from Invitrogen. Descriptions of the plasmids, pUC119[Pg] and pSVI7/SG[K2tPAP], containing the entire coding sequences of Pg and the K2tPA-protease region of tPA, respectively, were provided earlier (Whitefleet-Smith et al., 1989; Nilsen et al., 1997). Construction of the Fusion Protein Expression Plasmids. The cDNAs of K2tPAP and plasminogen were used as templates for PCR-based mutagenesis. The first step consisted of construction of the tandem K1Pg-IEGR-K2tPA fusion polypeptide. For this, two pairs of oligonucleotides were synthesized that contained restriction endonuclease sites for AVrII site at the 5′-end and NotI at the 3′-terminal region of the final cDNA construct. The first pair of PCR primers was used with pUC118-Pg as the template.
7654 Biochemistry, Vol. 36, No. 25, 1997 1. 5′ forward primer, 5′-GAT C′CT AGG TCA GAG TGC AAG ACT GGG (AVrII site underlined). After cleavage by AVrII and insertion of the product into the same restriction site of the P. pastoris transfer vector, pPIC9K, this primer encodes the sequence PRSEC1KTG. The sequence is preceded by the AVrII recognition site. The first two amino acids restore the 3′-end of the polylinker site of the transfer vector, and the next two amino acids flank the first C residue of K1Pg. The next three amino acids are residues 2-4 of K1Pg. 2. 3′ reverse primer, 5′-TGC ATT GGA tct tcc ttc gat ATC ACA CTC AAG A (the fXa cleavage site, IEGR, is in lower case letters). This primer encodes the sequence LEC79DIEGRSNA, where LEC79 represents the COOH terminus of K1Pg, followed by DIEGR′SNA (the fXa-sensitive cleavage site which is excised at R′ by the enzyme). The second set of PCR primers was employed with plasmid pSVI7/SG[K2tPAP]. 3. 5′-GGA AGA TCC AAT GCA GAC TGC TAC TTT GGG. The first 15 bases (italicized) of this forward PCR primer overlap those of primer 2, above, and encode the sequence GRSNA. The first two amino acids originate from the carboxy terminus of the fXa-sensitive cleavage site (GR). The next three (SNA) begin the construction of K2tPA. These are then followed by amino acids at the amino terminus of K2tPA, Viz., DC1YFG. 4. 5′-AT GC′GGCCGC TTA CTA GGA GCA GGA GG. This primer codes for the sequence PSC82S(stop)(stop), where PSC82 represents the carboxyl terminus of K2tPA, followed by two stop codons and the 3′ NotI site (underlined). The final PCR product was then digested with AvrII/NotI and cloned into the polylinker site of pPIC9K through these same restriction sites. Oligonucleotide-Directed Mutagenesis. The QuickChange protocol (Stratagene) was employed to alter the codon for W63 of K2tPA to those that individually code for H, F, S, and Y. The sequences of the pairs of oligonucleotides used in this procedure were as follows (the codons used for the W63 mutants are in lower case letters):
The plasmid containing the K1Pg-IEGR-K2tPA coding sequence in pPIC9K was amplified from Escherichia coli strain DH5R. This plasmid was thermally denatured and reannealed to the above oligonucleotide primer pairs containing the desired mutagenic nucleotides. Then, using the nonstrand-displacing action of Pfu DNA polymerase, the primer was extended with incorporation of the mutagenic primers, a process that resulted in nicked circular DNA strands. The nonmutated parental DNA template, which contained methylated G (Gm6), was digested with DpnI
Chang et al. (recognition sequence 5′-Gm6ATC-3′). The circular, nicked double-stranded DNA was then transferred into XL2-Blue competent cells, which repaired the nicks in the mutated plasmid. The final product was amplified in XL2-Blue and subjected to complete nucleotide sequencing to confirm the mutations and the integrity of the entire constructs. Expression and Purification of the Fusion Polypeptide K1Pg-IEGR-K2tPA. The plasmid pPIC9K[K1Pg-IEGR-K2tPA] and the mutant constructions were linearized with the restriction enzymes BglII, SalI, or SacI and used for transformation of P. pastoris strains GS115 or KM71 by means of electroporation. A two-step procedure was employed for selection of transformants that were His+ and G418 resistant. First, after transformation, the cells were plated on His- plates and were allowed to grow for 2-5 days. Colonies were selected from these plates and transferred to nutritionally rich plates with variable concentrations (0.25, 2.0, and 4.0 mg/mL) of G418. The colonies that grew on 4 mg/mL G418 were considered as positive transformants with multiple copies of G418-resistant genes and target genes. The multicopy His+/GS115 transformants were selected for the Muts (methanol utilization slow) and Mut+ (methanol utilization plus) phenotypes as described in the P. pastoris expression kit manual, version C (Invitrogen). The His+/ KM71 transformants are naturally of the Muts phenotype (because of interruption of the AOX1 gene by that of ARG4), and further determination of the methanol growth properties was not necessary. High biomass fermentation of the selected yeast clones (Mut+ colonies were used for large-scale fermentations whenever possible) was accomplished with a BioFlo 3000 bench-top fermentor (New Brunswick Scientific, Edison, NJ). The inoculum for the fermentation was obtained from shaker flasks containing 200 mL of phosphate-buffered YNB (yeast nitrogen base)/1% (v/v) glycerol and was grown overnight at 30 °C to an OD of 2-6 absorbance units. The details of our fermentation protocols have been published earlier (Nilsen et al., 1997). The only differences in the procedures between wtr-K2tPA and the mutants involved the time of fermentation. While the wild-type polypeptide construct was allowed to ferment for 48-60 h after methanol induction, some of the mutants, Viz., K1Pg-IEGR-[W63H]K2tPA and K1PgIEGR-[W63S]K2tPA, were found to be degraded in the K2tPA portions under these conditions. When this occurred, shorter fermentation times of 14 h after methanol induction were used. Despite the resulting lower yields of polypeptide, the quality of the material obtained was very high. Purifications of the fusion polypeptides were achieved by affinity chromatography. Specifically, the supernatant was adjusted to pH 7.4 with NaOH and loaded onto a lysineSepharose column, equilibrated and washed with a buffer containing 50 mM Tris-HCl/50 mM NaCl, pH 7.7. The fusion polypeptides were adsorbed to the column by nature of the specific affinity of K1Pg (and in some cases also through the K2tPA portion of the construct) and were then eluted after adjusting the column wash buffer to 0.2 M EACA. The purified polypeptides were then dialyzed against H2O and lyophilized. BoVine fXa-Catalyzed Digestion of the Fusion Polypeptides and Isolation of the K2tPA Region. The K1Pg-IEGR-K2tPA constructs were mixed with bovine fXa at a peptide:enzyme
Tryptophan-63 of tPA Kringle 2 ratio of 50:1 (w:w) and allowed to incubate at 37 °C for 3 h. The digest was then applied to a lysine-Sepharose column and washed with a solution containing 50 mM Tris-HCl/50 mM NaCl, pH 7.7. After the baseline absorbance (280 nm) decreased to
